www.ietdl.

org
Published in IET Renewable Power Generation Received on 2nd September 2008 Revised on 21st April 2009 doi: 10.1049/iet-rpg.2008.0080

Special Issue selected papers from EWEC 2008

ISSN 1752-1416

Comparison of integration solutions for wind power in the Netherlands

B.C. Ummels1 E. Pelgrum2 M. Gibescu3 W.L. Kling4
Technical Project Management, Offshore, Siemens Wind Power, Prinses Beatrixlaan 800, 2595 BN Den Haag, The Netherlands 2 Market and Regulation Department, TenneT TSO, Utrechtseweg 310, 6812 AR Arnhem, The Netherlands 3 Power Systems Laboratory, TU Delft, Mekelweg 4, 2628 CD Delft, The Netherlands 4 Power Systems Laboratory, TU Delft, Asset Management Department, TenneT TSO, Mekelweg 4, 2628 CD Delft, The Netherlands E-mail: bart.ummels@siemens.com
1

Abstract: In this study, a commercially available unit commitment and economic despatch (UC ED) tool is extended for the simulation of wind power integration in an international environment. An existing generation unit database for the Netherlands is extended to include conventional generation portfolios of neighbouring areas to the Netherlands. Furthermore, wind power in Germany is modelled such that the spatial correlation between wind speeds at different locations in the Netherlands and Germany is maintained. These additions allow the assessment of the benets of international exchange for wind power integration and a comparison with other integration solutions. The UC ED tool is applied for annual simulations of a power system with generation portfolios foreseen for the year 2014. Four variants for international exchange possibilities are investigated for different wind power penetrations. The opportunities of the following integration solutions are assessed: use of conventional generation in isolated systems,use of international markets, exible combined heat and power (CHP), pumped hydro energy storage, compressed air energy storage and interconnection to a hydro-based system. The solutions are placed in an order of potential with respect to technical, economical and environmental aspects. The results show that the advantages of international exchange for wind power integration are large and provide an alternative for the development of energy storage facilities.

Introduction

Unit commitment and economic despatch (UC ED) are two optimisation tasks requiring different optimisation procedures and comprising different time frames. UC decisions are typically assessed only once or twice a day, whereas generation despatch is carried out throughout the day. With the reasonable predictability of system load, intra-day calculations for UC are in principle necessary only when unexpected, signicant changes occur in generation (e.g. outages) or demand. The emergence of international markets and the growth of wind power has complicated the optimisation of UC ED in the sense that more variables and uncertainties (i.e. market prices, wind power forecasts) must be taken into account. Since there is only a conceptual IET Renew. Power Gener., 2009, Vol. 3, Iss. 3, pp. 279 292 doi: 10.1049/iet-rpg.2008.0080

difference between markets and the traditional generation scheduling (i.e. market participant price bids instead of operating cost minimisation), solutions for the traditional central optimisation of UC ED based on cost are still highly relevant [1]. A wide range of models has been developed for the simulation of UCED or electricity market operation ranging from weekly operations planning to generating unit investment planning. The Danish SIVAEL model [2] is a UCED tool capable of minimising total system cost while supplying local heat and power demand. This model has been applied in [3] for system integration studies of large-scale wind power in Denmark. The Wilmar model [4] is used to simulate alternative solutions for the integration of large-scale wind power into 279

& The Institution of Engineering and Technology 2009

Authorized licensed use limited to: Edinburgh University. Downloaded on January 11, 2010 at 06:55 from IEEE Xplore. Restrictions apply.

www.ietdl.org
interconnected power systems providing input to decision makers. The model has been applied for the exploration of different integration solutions for wind power, for example heat boilers for making combined heat and power (CHP) plants more exible [5] or the use of compressed air energy storage (CAES) [6]. The use of linear programming however limits its application for the modelling of large systems or international studies because calculation times increase exponentially. The model used in this paper is PowrSym3, a multi-area, multi-fuel, chronological generation cost simulation model for electrical power systems including CHP systems, energy storage and energy limited fuel contracts [7]. An essential advantage of this tool is its ability to include large numbers (  100) of generation units, multiple areas with interconnectors and wind parks. In earlier research [8], the tool was applied for an isolated system. It was shown that the rst integration limit for large-scale wind power in the Netherlands is the so-called minimum-load limit: must-run generation on minimum output and high wind power production threaten to exceed load during off-peak periods. Energy storage may provide opportunities for the integration of additional wind power during such periods [9] but was found to be disadvantageous due to very high investment cost and increased CO2 levels on moderate wind power levels. In both articles, it was found that international exchange is a key parameter in determining the technical limits that may exist for wind power integration. However, international exchange will be increasingly be inuenced by the wind power levels and outputs in different countries, requiring that correlations between wind power output in these countries are correctly incorporated. This paper presents the results of chronological power system simulations of large-scale on- and offshore wind power in an international system. The geographical correlations of wind power in different countries (in this case, the Netherlands and Germany) and of the load are specically taken into account. Models are developed for the conventional generation portfolio in neighbouring areas of the Netherlands and for the interconnectors between the different areas. The model is then applied to simulate different gate closure times of international markets in order to assess the possible benets of additional trade possibilities for incorporating the variability and limited predictability of the wind. For the rst time, the correlations of wind power are taken into account on an international level, allowing the assessment of integration possibilities for wind power by using international exchange in a much better way [10]. This paper is organised as follows. First, an overview of the earlier work and the UC ED model are presented. Simulation parameters are identied to assess the technical, economical and environmental aspects of integrating largescale wind power into power systems. Then, the system to 280 & The Institution of Engineering and Technology 2009 be simulated is specied in detail, comprising load, generation units, and wind power. Representative models of neighbouring countries of the Netherlands are included as well as transmission capacities between the different countries. Then, an overview is presented of all system simulations which have been performed and the simulation set-up. Different simulations are performed for several wind power penetration levels, taking into account the possibilities of international exchange and different integration solutions for wind power.

Simulation parameters

Since this paper is specically focused on the technical integration of wind power in electrical power systems, different technical limits for the system integration of wind power will be identied rst. Relevant simulation parameters will be identied in order to quantify these limits, if any. After this, the economical and environmental impacts of wind power on power system operation are discussed and simulation parameters for the assessment of these are also identied.

2.1 Technical limits

In the simulation of the chronological UC ED for the power system as a whole including wind power, technical limitations for the system integration of large-scale wind power are minimum-load problems, insufcient upward power reserves and insufcient downward regulation [8]. The rst technical limit for wind power integration is expected to concern minimum-load problems. This means that technical operating constraints of conventional units (mustrun status, minimum power level, minimum up-times) prevent a full integration of available wind power during low-load periods. Furthermore, signicantly larger amounts of power reserves will be required to balance wind power variations on a 15-min time scale on top of the existing load variations. Also, technical operating constraints, unscheduled outages or large wind power forecast errors may result in insufcient upward reserves. In case of insufcient upward reserves for matching simultaneous load increases and wind power decreases, the simulation results will show violations of spinning reserve requirements and ultimately, energy-not-served (ENS). When there is insufcient downward regulation, wind power will be ramped down as a last resort, resulting in a waste of available wind energy. A detailed analysis of the simulation results for the consecutive 15 min to 15 min steady-states is necessary to conrm this limit has indeed been reached, since wasted wind energy may also be related to minimumload issues.

2.2 Economical impacts

Wind power has a number of economical impacts on power system operation, which are all related to the low marginal cost of wind power. The extent to which wind power IET Renew. Power Gener., 2009, Vol. 3, Iss. 3, pp. 279 292 doi: 10.1049/iet-rpg.2008.0080

Authorized licensed use limited to: Edinburgh University. Downloaded on January 11, 2010 at 06:55 from IEEE Xplore. Restrictions apply.

www.ietdl.org
indeed lowers the short-term marginal cost curve of the system as a whole, depends on a number of factors. The most important are the technical exibility of the system wind power is integrated into and the extent to which the market allows an efcient exploitation of this exibility. The extent to which wind power may decrease market prices in a structural way depends on the cost of additional power reserves and the decreased operational efciencies of conventional units, which increase total operating cost. In this paper, these aspects are taken into account by the simulation model and are expressed in the total system operating cost. The operating cost [ME/year] will be monitored for the Netherlands specically. This also applies for the total electricity produced [TWh/y] per conventional generation technology (nuclear, coal, Combined Cycle Gas Turbine (CCGT), CHP, gas turbine etc.).

3.1.1 New conventional generation: New coal-red units (total 5.3 GW) are assumed to have a higher maximum operating efciency of 44.5% and an efciency curve shape similar to coal-red units already present in the database. These new units do not have a must-run status but a minimum up-time and down-time of 16 h. This allows for temporary shut-downs during periods of low prices, for instance during weekends.
New natural gas-red units (total 3.68 GW) are CCGTs with a maximum operating efciency of 58% and an efciency curve shape similiar to existing CCGTs. Of the new CCGTs, two units (1.26 GW total) are modelled as industrial CHPs with a must-run status, delivering steam to a separate industrial heat area. The heat load curves of these areas are modelled based on the existing curves in the database. The heat areas are equipped with heat boilers which are assumed only to be used during maintenance of the CHP units, similar to existing CHP unit models.

2.3 Environmental impacts

The environmental aspects of wind power are the most important driving force behind the development of wind power. Electricity generated by wind power replaces fossilred conventional generation and thereby saves fuel and CO2 and other emissions. The simulation model integrally takes into account emission cost during the optimisation of the UC ED. For monitoring the environmental impacts of wind power, emissions (CO2 , but also NOx and SO2) will be assessed at the system level for the Netherlands and presented for each conventional generating technology.

3.1.2 New distributed generation excluding wind power: New distributed generation (DG) capacity
in the Netherlands mostly involves installations, that is gas engines, in Dutch greenhouses. These CHP units produce electricity, heat and CO2 for the greenhouses, with heat boilers as back-up and heat buffers for several days of storage. Due to the availability of a full heat back-up, the generation units are operated against spot market prices and have a very high operational exibility. A total capacity of 3000 MW of gas engines has been modelled, with maximum electrical efciency of 40%, no minimum uptime or down-time and a 10% unavailability. Ramp rates of these DG units are estimated to allow a ramp from minimum to maximum output within 15 min. Other, existing DG, is modelled as nondespatchable capacity (3400 MW) and aggregated into a xed schedule, simulated as must-take power on the basis of natural gasred generation. This capacity represents nondespatchable industrial units, waste incineration and other small DG units. The output of this DG is assumed to be 50% constant and 50% variable with system load and has an efciency of 19%.

3 Power system model specication

The power system model comprises a physical representation of different areas, each representing the generating systems of the Netherlands and its neighbouring areas: Belgium, France, Germany, Norway and Great Britain. Interconnections between these areas are modelled explicitly, transmission constraints within each area are not considered here.

3.1 The Netherlands

The existing models database contained models for all larger ( 60 MW) conventional generation units in the Netherlands. This database is updated to represent the Dutch power system for 2014, the year chosen for investigation. Up to 2014, a large amount of new conventional generating capacity is planned or under development, comprising especially new coal- and gas-red generation capacity. It is foreseen that some of the older installations will be shut down by that time but market parties have not provided information on this. The Dutch generating portfolio for 2014 is based on the existing portfolio with the addition of new units specied in [11]. The interconnection capacity of the Netherlands is based on transmission capacity forecasts of TenneT TSO for 2014 [11] and ETSO [12]. IET Renew. Power Gener., 2009, Vol. 3, Iss. 3, pp. 279 292 doi: 10.1049/iet-rpg.2008.0080

3.1.3 Wind power: For the Netherlands, a 1-year data series of 15 min. wind power data for seven wind power penetration levels have been developed for 0 GW, 2 GW (225 MW offshore), 4 GW (1 GW offshore), 6 GW (2 GW offshore), 8 GW (4 GW offshore), 10 GW (6 GW offshore) and 12 GW (8 GW offshore), based on the methodology described in [13]. It is assumed that wind power does not replace any conventional capacity, allowing an accurate comparison of the technical, economical and environmental impacts of wind power between different penetration levels. In practice, wind power has a certain capacity credit [14, 15] and will lead to a change in the total installed generation capacity, but this is not considered here.
281

& The Institution of Engineering and Technology 2009

Authorized licensed use limited to: Edinburgh University. Downloaded on January 11, 2010 at 06:55 from IEEE Xplore. Restrictions apply.

www.ietdl.org
3.2 Neighbouring areas
3.2.1 Areas and interconnectors: In this paper, the Netherlands (NL) is assumed to be a central area with interconnections to Belgium (B), France (F), Germany (D), and Great Britain (GB), and Norway (NOR) to a very limited extent. Each country is represented as a single area comprising generation and load with interconnections to neighbouring areas based on [11, 12]. Only cross-border transmission capacities between countries are taken into account. The power system with all areas and interconnectors developed and used in this paper is shown in Fig. 1, representing the Netherlands as part of the West-European interconnected system. The NODE-area (grey) is an empty area used to incorporate the transmission capacity limits foreseen by TenneT TSO for 2014 for the Netherlands with Belgium and Germany together. Apart from the separate interconnection capacities which may exist between the Netherlands and Belgium (2300 MW) and Germany (5400 MW), a net export/import transmission capacity maximum of 5650 MW is applied between the Netherlands and these countries.
The transmission capacities NL NOR (NorNed), NL GB (BritNed) and F GB (Cross-Channel) are high voltage direct current (DC) sub-marine interconnectors, for which an availability of 98% is assumed, other interconnections have an availability of 100%. Transmission losses are assumed to be 5% of the transmitted power for NorNed, and 4% for BritNed and Cross-Channel. using annual growth rates up until 2014 based on [17, 18]. The growth rates for load in 2014 relative to 2007 are 1.08, 1.08, 1.03 and 1.10 for Belgium, France, Germany and Great Britain, respectively.

3.2.3 Fossil-red generation: Models for conventional generation units in the neighbouring areas of the Netherlands were developed based on installed capacity estimates made in [17, 19]. Generating technology efciencies and other technical factors were estimated based on and using the existing models for the Dutch units in the database. Each generation technology type outside the Netherlands is modelled as an aggregation of units with identical characteristics, with a total generating capacity equalling the capacity foreseen to be installed in 2014. Nuclear generation units in Germany and Great Britain are modelled as having a technical full-load must-run operational status and no ramp rate. French nuclear units have only a must-run status, which allows part-load operation during moments of low load. 3.2.4 Hydro power and pumped hydro: Total annually available hydro energy in Germany, France and Great Britain are estimated based on yearly statistics from [20]. Norway is modelled only as a pumped hydro unit without inow. Since PowrSym3 optimises the UC ED on a weekly basis, an annual optimisation of hydro energy per week must be done rst. It is assumed that reservoir hydro is operated such that the same amount of energy is available for each week. The UC ED optimises the hourly distribution of this weekly energy during the week itself. Hydro power has a very high operational exibility: ramp rates are assumed to be sufcient to allow a full ramp between start and maximum output within 15 min. Pumped hydro units are modelled with a similar exibility as hydro power, but without a weekly energy inow. The UC ED of pumped hydro in each area is optimised on a weekly basis based on temporal differences in marginal cost in that area. The generating and pumping efciency are estimated at 90% each. The average unavailability of pumped hydro is determined at 2%, due to the absence of thermodynamic processes in these units. 3.2.5 Wind power: Wind power in neighbouring areas is only taken into account for Germany. This is because Germany already has a large installed capacity of wind power, which is foreseen to increase signicantly [17, 21]. Furthermore, wind power outputs in the Netherlands and Germany are strongly correlated and a large interconnection capacity is available between these countries. In a NorthWest European market, the presence of large-scale wind power in Germany could present additional barriers for wind power in the Netherlands.
In order to correctly incorporate the correlation between wind power in the Netherlands and Germany, 15-min average wind power data for the German areas EON-Netz, RWE and Vattenfall were obtained from the respective TSOs for the same period as the Dutch meteorological IET Renew. Power Gener., 2009, Vol. 3, Iss. 3, pp. 279 292 doi: 10.1049/iet-rpg.2008.0080

3.2.2 Area load: Load data for the areas outside the Netherlands for the year 2014 are developed using UCTE load data for the relevant countries for the year 2007 [16]. This assures that correlations among momentary loads in all countries are automatically taken into account. The load data are processed similarly as with the Dutch load data

Figure 1 Areas and interconnections included in the simulation model 282 & The Institution of Engineering and Technology 2009

Authorized licensed use limited to: Edinburgh University. Downloaded on January 11, 2010 at 06:55 from IEEE Xplore. Restrictions apply.

www.ietdl.org
data (1 June 2004 to 31 May 2005). Also, day-ahead forecast data were obtained for this period. Wind power and wind power prediction data are scaled to represent the 32 GW of installed capacity foreseen for 2014. It is found that the correlation between the 15 min aggregated average wind power data sets for the Netherlands and Germany is 0.73. Thus, interconnection capacity to Germany may not be (fully) available for exports during moments of low-load and high-wind power, resulting in wasted wind energy. Wind power in Germany is modelled such that it is integrated with a higher priority than Dutch wind power in order to guarantee that wind power in the Netherlands is not integrated into the system at the expense of German wind power. only on the differences between marginal cost in the Netherlands and the value of the energy contained in the Norwegian reservoir. This approach is sufcient for this research, with its focus on the system integration of wind power rather than the economical benets of interconnection capacity, and makes it unnecessary to model the Scandinavian system explicitly.

System simulations

3.3 Power system overview

Installed generating capacities per technology are based on forecasts made for the relevant countries for 2014 in [11, 17, 19]. The installed capacities are presented per technology and per country in Table 1. Norway is modelled as a pumped hydro unit with a generating/pumping capacity equalling the interconnection capacity between the Netherlands and Norway. The hydro power system of Norway has a high operational exibility and is assumed to be available for imports from and exports to the Netherlands at all times. The exchange is dependent Table 1 Generation technologies per country in 2014 Technology nuclear coal lignite CCGT CHP Ind. BF Gas Ind. CCGT CHP Res. CCGT gas turbine oil reservoir hydro pumped hydro RoR hydro other total wind power maximum load demand (TWh/y)

In the system simulations performed here, the UC ED is optimised on a central basis: it is assumed that electricity markets function well. The objective function is formulated at the system level that is no other transmission constraints are taken into account other than those specied between different areas. The UC ED is calculated using the equal marginal cost method, in which the objective function is the total cost for heat and power generation, including emission cost. Decremental despatch and de-commitment cost are calculated for all units included in the simulation. The simulation program calculates an optimal maintenance schedule for the simulated year beforehand and determines unscheduled outages using Monte Carlo for all generation units, energy storage units and heat boilers. The commitment and despatch of energy storage and heat boilers is based on the minimisation of the overall operating cost of the system.

The Netherlands (GW) Belgium (GW) France (GW) Germany (GW) GB (GW) Norway (GW) 0.4 9.5 2 4.0 0.9 1.5 7.1 0.6 2 2 2 2 6.3 30.4 10.0 21.0 126 5.9 2.6 2 2 2 2 5.0 1.5 2 2 1.3 0.1 0.4 16.8 2 15.2 97 64.9 6.0 2 2 2 2 4.0 1.1 9.2 13.6 4.2 7.9 2 110.9 2 87.1 518 14.1 32.0 18.9 2 2 2 15.1 4.0 5.3 3.7 5.5 2 8.2 106.8 32.0 80.5 550 11.9 30.4 2 2 2 2 24.4 7.0 8.4 1.8 3.0 2 2 86.9 2 65.5 367 2 2 2 2 2 2 2 2 2 2 0.7 2 2 0.7 2 2 2

IET Renew. Power Gener., 2009, Vol. 3, Iss. 3, pp. 279 292 doi: 10.1049/iet-rpg.2008.0080

283

& The Institution of Engineering and Technology 2009

Authorized licensed use limited to: Edinburgh University. Downloaded on January 11, 2010 at 06:55 from IEEE Xplore. Restrictions apply.

www.ietdl.org
As an illustration of the system simulations performed in this paper, Fig. 2 provides an overview of the UC ED in the Netherlands during 1 week for the scenario with 12GW wind power. The graph shows generation levels for DG, thermal units, integrated wind power and the amount of wasted wind energy. Total generation by conventional thermal generation units follows the system load, DG and wind power. In this particular week, wind power is ramped down at moments of high wind power and low load (all nights, except Sunday when there is little wind power available) to prevent minimum-load problems. A good example of the use of thermal generation for balancing the combined variations of load and wind power can be seen on early Sunday morning (thermal generation ramps up and wind power is decreasing). 2. International market gate closure time day ahead 3. International market gate closure time 3 h ahead 4. International market gate closure time 1 h ahead The base-case for the simulations is the Netherlands seen as an isolated power system. International exchange with Belgium, France, Germany, Norway and Great Britain is assumed to be zero at all times. This variant serves as a reference to consider the integration of wind power in the Dutch power system using the technical capacities available in the Netherlands only. The other market designs all comprise international exchange possibilities between the Netherlands and its neighbours, but using different gate closure times. This means that the imports and exports of the Netherlands are optimised using the wind power forecast available at market gate closure. After market gate closure, the international exchange schedules become xed and are executed as scheduled exchanges during actual operation. For the day-ahead market closure, wind power forecast errors are signicant [8]. This will result in a suboptimal scheduling of imports and exports from a wind power integration point of view. Forecast errors will have decreased by about 50% if market gate closure is delayed up to 3 h ahead of operation, and no forecast errors are assumed to be present for a market design with near realtime operation (1 h ahead), which allows an optimal scheduling of international exchange considering wind power.

4.1 Base variants

The base simulation variants consider seven levels for wind power capacity installed in the Netherlands, four designs of international markets and three wind power forecast methods. The base variants will be used to quantify the technical, economical and environmental impacts of wind power. For all base-variants, it is assumed that wind power is integrated into the system by taking into account wind power in the optimisation of the UC ED of conventional generation capacity. In case no international exchange market is available, only the Dutch conventional generation units are used.

4.1.1 Wind power levels: All base simulations will be carried out for six wind power penetration levels, as mentioned earlier (2 12 GW), and for a 0-MW wind power variant to be used as a reference. 4.1.2 International exchange: International exchange is modelled and simulated for four market designs:
1. No international exchange

4.2 Wind power integration solutions

Technical limits may exist for the system integration of wind power in the Dutch power system. After the determination and quantication of these limits, different alternatives will be explored for overcoming these integration limits. The solutions considered here include three energy storage options for the Netherlands (PAC, UPAC and CAES), more exible industrial CHP units and an increased

Figure 2 Example of a UC ED for 1 week in the Netherlands for 12 GW installed wind power capacity and no international exchange 284 & The Institution of Engineering and Technology 2009 IET Renew. Power Gener., 2009, Vol. 3, Iss. 3, pp. 279 292 doi: 10.1049/iet-rpg.2008.0080

Authorized licensed use limited to: Edinburgh University. Downloaded on January 11, 2010 at 06:55 from IEEE Xplore. Restrictions apply.

www.ietdl.org
interconnection capacity to Norway. Models for these alternatives were developed in [9]. For all simulations, conventional generation units are used for balancing wind power. More exible base-load generation capacity may provide additional technical space for wind power during low-load, high-wind situations. For this alternative, the commitment status of selected industrial CHP units is changed from must-run to economic. The existing heat boilers could in principle take over the generation of steam for the industrial processes at times when the CHP is shut down to allow a further integration of wind power. The energy storage alternatives included in this research include surface pumped accumulation energy storage (PAC), underground PAC (UPAC), CAES and creating additional interconnection capacity between the Netherlands and the hydro power-dominated system of Norway (NN). It can be noted that no specic attempt has been made here to optimise the design of the energy storage: the energy storage capacities and reservoir sizes applied in [22] have been adopted here.

4.3.2 Fuel and emission cost: Fuel and emission cost have been determined based on price forecasts stated in [23] for the year 2015. The prices for coal, lignite, gas, oil, uranium and CO2 used in this paper are 2.00 E/GJ, 1.36 E/GJ, 5.00 E/GJ, 10.50 E/GJ, 1.00 E/GJ and 20.00 E/ton, respectively. Emission cost are included in the calculation of the marginal operating cost of each thermal generation unit. 4.3.3 Overview: In Table 2, an overview is shown of all simulations performed in this paper. Simulations are performed for different wind power penetrations (seven WP scenarios), international market designs (NL no international exchange, 24 h international market gate closure at 24 h ahead, 3 h idem at 3 h ahead, 1 h idem at 1 h ahead). For all simulations, a best available wind power forecast is used (i.e. a rolling forecast). Furthermore, solutions for wind power integration are explored (CHP exible CHP units, ES energy storage, consisting of the alternatives PAC, UPAC, CAES and NN2 (NN2 is only simulated for the 1 h ahead gate closure market design)). A total number of 63 simulations (7.(4 5)) are carried out.

4.3 Simulations set-up

The simulations of the UC ED are carried out for a future year 2014, with a resolution of 15 min, for different installed wind power capacities, wind power forecasts, international market designs and balancing solutions. An optimised unit maintenance schedule is calculated ahead of each simulation and unscheduled outages are introduced using the Random Monte Carlo method for all generation units, energy storage units and heat boilers, for every week. UC ED are centrally optimised (well functioning electricity markets) in order to achieve minimum operating cost at the system level, while all technical constraints are fullled. The UC ED is calculated using an equal marginal cost method, in which the objective function is the total cost for generating both heat and power. A calculation of unit despatch is performed every 15 min using the given load prole and an estimation of the wind power production levels.

5 Impacts of wind power on power system operation

5.1 Technical limits
Possible technical limits for the integration of wind power are minimum-load problems and insufcient upward power reserves or downward regulation reserves for balancing load and wind power variations. The simulation results for all base simulation variants (012 GW wind power) do not report ENS nor spinning reserve violations in the Netherlands. From this result, it can be concluded that sufcient upward power reserves and downward regulation are present at all Table 2 Overview of the UC ED simulations WP NL 24 h 3 h 1 h 0 2 4 6 8 10 12 BF BF BF BF BF BF BF BF BF BF BF BF BF BF BF BF BF BF BF BF BF BF BF BF BF BF BF BF CHP ES

NL-BF NL-BF NL-BF NL-BF NL-BF NL-BF NL-BF NL-BF NL-BF NL-BF NL-BF NL-BF NL-BF NL-BF

4.3.1 Power reserves: Spinning reserves and fast regulating power are provided by the non-despatched capacity of committed generation units. All coal- and gasred generation units in the Netherlands sized 60 MW and above are assigned with a spinning reserve contribution of 1% of its nominal power. Furthermore, some capacity must be reserved for regulating power. The needed amounts of reserves to guarantee sufcient capacity for load following is determined during the optimisation of the UC ED based on load and wind power forecasts. On top of the reserves determined by PowrSym3, additional requirements for reserves may be imposed. For the Netherlands, an extra spinning reserve of 1600 MW (twice the largest single generator) is assumed, for the other areas except Norway the reserve requirement is set at 2000 MW.
IET Renew. Power Gener., 2009, Vol. 3, Iss. 3, pp. 279 292 doi: 10.1049/iet-rpg.2008.0080

WP wind power, NL isolated Dutch system, 24 h 24 h ahead international market gate closure, CHP exible CHPunits, ES energy storage, comprising PAC, UPAC, CAES and NN2, BF best available wind forecast 285

& The Institution of Engineering and Technology 2009

Authorized licensed use limited to: Edinburgh University. Downloaded on January 11, 2010 at 06:55 from IEEE Xplore. Restrictions apply.

www.ietdl.org
moments during the year in order to balance the aggregated load and wind power and load variations. This was to be expected since the total generation portfolio is rather large compared to the maximum load. The integration of wind power in the Dutch power system does however result in an increase of heat production at CHP locations (especially residential), wasting of wind energy (especially during low-load periods) and increased exports to neighbouring countries (idem). This indicates that minimum-load problems pose a technical limit for the integration of wind power, which is in line with the observations made in [8]. The simulation results for wasted wind energy and international exchanges vary considerably between the different international market designs, since they are mutually dependent, and to some extent between the different wind power forecast methods investigated here. use of international exchange provides signicant additional space for the integration of wind power. Fig. 4 focuses further on a comparison of the amount of wasted wind energy for different market designs. Only wind power forecast errors in the Netherlands are considered here. In case no interconnection capacity is available, an estimated amount of 6.2 TWh/y or 15% of available wind energy in the Netherlands cannot be integrated into the system. In case international exchanges can be used for exports at high wind power levels, additional wind power can be integrated, with only 0.05 TWh or 0.1% of available wind energy being wasted for the 1 h. ahead market gate closure. Interestingly, a day-ahead or 3 h ahead international market gate closure time results in larger amounts of wasted wind power at smaller wind power capacities. This is the result of the methodology applied for the optimisation of international exchange at market gate closure, which is based on the assumption that all feasible international transactions are being made. In case a signicant wind power forecast error is present at the moment that these transactions become xed, scheduled imports may prevent the integration of unpredicted surpluses of wind power, leading to larger amounts of wasted wind energy. For large wind power penetrations, however, the benets of international exchange capacity outweigh the disadvantage of forecast errors. Clearly, a more conservative

5.1.1 Wasted wind energy: In Fig. 3, wind energy integrated into the Dutch power system is shown for different wind power penetrations and different market designs. Wasted wind energy becomes signicant in the range of 68 GW installed wind power capacity for the Dutch power system, in the market design without international exchange. The slight change in steepness of the available wind energy curve at 2 and 6 GW installed capacity is due to increased capacity factor of wind power (offshore against onshore). The

Figure 3 Integrated and wasted wind energy in the Netherlands

Figure 4 Wasted wind energy for different international market designs 286 & The Institution of Engineering and Technology 2009 IET Renew. Power Gener., 2009, Vol. 3, Iss. 3, pp. 279 292 doi: 10.1049/iet-rpg.2008.0080

Authorized licensed use limited to: Edinburgh University. Downloaded on January 11, 2010 at 06:55 from IEEE Xplore. Restrictions apply.

www.ietdl.org
scheduling of international exchanges (imports) will result in less wasted wind energy. This result illustrates the benets of postponed international market gate closure times for integrating wind power. In case interconnection capacity is available and the market design allows an adjustment of international exchange up until the moment of operation (1 h ahead international market gate closure), the potential for additionally integrated wind energy is high. Still, even the most exible international market design cannot prevent a small amount of wasted wind energy, starting from 8 to 10 GW installed capacity in the Netherlands (bottom of Fig. 4, not visible in Fig. 3). The reason for this is that, even though international transmission capacity may be sufcient, this capacity is not always fully available for exports. This applies to Germany in particular. Germany has a signicant must-run conventional generation capacity and a large amount of wind power (32 GW in the year 2014) which is highly correlated (0.73) to wind power in the Netherlands. Both factors reduce the possibilities for exports of wind power from the Netherlands, especially during critical periods. In case no international exchange is possible for exports of excess wind power, the relative cost savings gained from wind power start to decrease from 8 GW installed capacity onwards. Limits in the operational exibility of conventional plants lead to sub-optimal despatch, reduced operating efciencies and, ultimately, increased wasting of available wind resources. In case the Netherlands is part of an international North-West European market, the technical integration limits for wind power are smallerand operating cost savings are higher than for an isolated Dutch system. In such an international environment, slightly over one half of the total economical benet of wind power is realised in the Netherlands, the rest is realised in neighboring areas.

5.2 Economical impacts

5.2.1 Operating cost: Fig. 5 shows the annual savings in
operating cost due to increasing wind power in the Netherlands, for the Netherlands itself without international exchange, and in case international exchange is possible (1 h ahead market gate closure time) for the North-West European system as a whole and the Dutch part as a dotted line. As the gure shows, the operating cost savings by wind power increase with the amount of wind power installed. For the fuel and operating cost used here, the overall annual operating cost savings by wind power are estimated to be in the order of 2 billion E annually for 12 GW wind power capacity. The higher cost savings for the Netherlands without international exchange are due to the higher marginal cost in the isolated Dutch system. Older and less efcient units will be the rst to be taken out of operation in an international market. Thus, the base-case with 0-MW wind power is already different with respect to marginal cost.

5.2.2 International exchange: In case international exchange is possible, the integration of wind power in the Netherlands inuences in principle the exchanges between all countries. In Fig. 6, imports and exports are shown for each country with each bar representing a wind power penetration scenario. Clearly, the Netherlands increases its annual exports and decreases its imports in case more wind power is installed. This inuences mainly imports and exports of Germany and Great Britain, and Belgium to a limited extent.
Large interconnection capacities are present between Germany and the Netherlands and Dutch wind power mainly decrease the full-load hours for base-load coal and lignite in Germany, but also some CCGT. Wind power furthermore reduces the exports of base-load coal power from Belgium and to a lesser extent from France during periods of low load (nights and weekends). Germany reduces its imports from France at times of high wind in the Netherlands. Exchanges with Norway stay constant in volume since it is modelled as such, although the moments of exports and imports are increasingly determined by wind power as its installed capacity in the Netherlands increases.

5.2.3 Generation output mix: In Fig. 7, the change in annual electricity output between different generation technologies is shown for the Netherlands (no international exchange) with increasing wind power capacity. Nuclear,

Figure 5 Annual operating cost savings by wind power IET Renew. Power Gener., 2009, Vol. 3, Iss. 3, pp. 279 292 doi: 10.1049/iet-rpg.2008.0080 287

& The Institution of Engineering and Technology 2009

Authorized licensed use limited to: Edinburgh University. Downloaded on January 11, 2010 at 06:55 from IEEE Xplore. Restrictions apply.

www.ietdl.org

Figure 6 International exchange in North-West Europe for 0 12 GW wind power installed capacity in the Netherlands

Figure 7 Absolute electricity production change and relative output per technology in the Netherlands for different wind power penetration scenarios, no international exchange being a full-load must-run technology, is not affected by wind power integration. Wind power decreases the full-load hour equivalents of especially coal-red units, but also CCGT CHP and CCGT are inuenced (note the large change at 2-GW wind power capacity for CCGT). This means that the prots of these units decrease, especially for must-run baseload generating units during low-load periods. The technical exibility of coal, CCGT CHP and CCGT does not require additional operating hours of peak-load gas turbines for wind power integration. DG (greenhouse gas engines) decreases its operation hours only very slightly: the must-run part is xed, and the exible units produce heat and power during other periods, with the heat being stored. amounts of CO2 emissions. In Fig. 8, the annual emission savings are shown for the Netherlands without international exchange, and for the North-West European system as a whole (international exchange is possible in this case), with the Dutch part of that as a dotted line. Emission savings are estimated to lie around 35 Mton annually for 12 GW wind power, with higher savings for the isolated Dutch system. In case international exchange is possible, older, less efcient units have already been pushed out of the market at the 0 MW wind power, therefore, emission savings by wind power are lower. It can be noted that emission savings also positively impact operating cost, since CO2 emission savings are part of the total operating cost. The change in steepness of the curves at 2 and 6 GW installed wind power capacity is due to the higher capacity factor of offshore wind power. For the isolated Dutch system, there is a change at 8 GW wind power due to the increasing amounts of wasted wind energy. The results for IET Renew. Power Gener., 2009, Vol. 3, Iss. 3, pp. 279 292 doi: 10.1049/iet-rpg.2008.0080

5.3 Environmental impacts

5.3.1 CO2 emissions: The simulation results clearly show that wind power leads to a saving of signicant
288 & The Institution of Engineering and Technology 2009

Authorized licensed use limited to: Edinburgh University. Downloaded on January 11, 2010 at 06:55 from IEEE Xplore. Restrictions apply.

www.ietdl.org

Figure 8 CO2 emission savings by wind power emission savings for SO2 and NOx show similar trends as CO2 . Total annual emissions show an estimated decrease of 22 Mton and 40 kton for SO2 and NOx , respectively, for an isolated Dutch system.

6.1 Wasted wind energy

In Fig. 9, the amount of wasted wind energy for all ve options can be observed. Clearly, all options considered here reduce the amount of wind wasted in the Netherlands due to minimumload problems. Energy storage and heat boilers increase the exibility of the Dutch system and thereby enable larger amounts of wind energy to be integrated. An extra interconnection to Norway creates a virtual energy storage with the same effects. When considering an isolated Dutch system, PAC has the highest potential for reducing the amount of wasted wind. An extra interconnector to Norway would provide a similar potential for this, if it could be used as assumed here. However, none of the energy storage options is sufcient to prevent wasted wind energy altogether. In case international exchange is possible with a 1 h ahead market gate closure, wasted wind energy would have been reduced much more than using any of the energy storage options investigated here.

6 Integration solutions for wind power

The integration of wind power in the Dutch power system leads to savings in emissions and operational cost. The technical limit for wind power integration is mainly the minimum-load, leading to increasing amounts of wasted wind energy, whereas the variability and limited predictability of up to 12 GW wind power can be solved using conventional generation units. In this section, solutions for an even better integration of wind power in the Dutch power system will be explored. Models for the different solutions were developed in [9] and consist of pumped accumulation energy storage (PAC, 1700 MW), UPAC (1400 MW), CAES (1500 MW), exible CHP-units by installing and using heat boilers (CHP, 1500 MW) and an extra interconnection between the Netherlands and Norway (NN2, 1400 MW). Due to the lack of information on decommissioning and investment decisions in the Netherlands, it is assumed that energy storage and the additional interconnection capacity to Norway do not replace conventional generation capacity.

6.2 System operating cost

The simulation results in Fig. 10 show that the operating cost savings by energy storage and boilers are positively correlated with the amount of wind power installed. Energy storage in the Dutch system amounts to savings between E 0 million (PAC) and E 17 million (CAES) annually for 2-GW wind

Figure 9 Wasted wind with exible CHP units, energy storage options and extra interconnection to Norway, no international exchange IET Renew. Power Gener., 2009, Vol. 3, Iss. 3, pp. 279 292 doi: 10.1049/iet-rpg.2008.0080 289

& The Institution of Engineering and Technology 2009

Authorized licensed use limited to: Edinburgh University. Downloaded on January 11, 2010 at 06:55 from IEEE Xplore. Restrictions apply.

www.ietdl.org

Figure 10 Operating cost savings by exible CHP units, energy storage options and extra interconnection to Norway, no international exchange power (currently installed capacity), increasing to E 80 million annually (PAC) for 12-GW installed capacity. The annual economical benets of heat boilers in the Dutch system are estimated to be E 38 million annually for 12GW wind power. Comparing the energy storage options in the Netherlands itself it can be observed that UPAC and PAC allow the highest operating cost savings followed by CAES and making CHP units more exible. This can be explained by the fact that PAC has the highest maximum pumping capacity, increasing the opportunities for large-scale energy storage at the lowest cost, compared to UPAC. CAES has a relatively small reservoir that reduces possible synergies with large-scale wind power. At high wind power penetration levels, power generation by CAES is increasingly pushed out of the market by wind power because of its operating cost (CCGT). Heat boilers are not used until the rst minimum-load problems occur at about 6-GW installed wind power; from then on, the operational cost savings of this solutions increase rapidly. Of all options considered here, the operational cost savings by increasing the interconnection capacity to Norway are highest (20 110 ME annually). It can be noted that this is without considering the possible additional benets of connecting the Dutch thermal-power system to the hydro-power system of Norway itself.

6.3 CO2 emissions

In Fig. 8, it was shown that system CO2 emission levels are reduced with the integration of large-scale wind power. Fig. 11 shows the emission levels of CO2 for energy storage and exible CHP units compared to the base-case. Interestingly, the simulation results show that the application of energy storage in the Dutch system increase the systems total CO2 emissions for wind power levels below 8 GW. Additional emissions with energy storage are highest at low wind power penetrations for NN2 due to its intensive use (because of its very large reservoir capacity) and lie around 1 Mton/y. As discussed previously in [9], from a CO2 perspective, energy storage is an environmentally friendly option only for very high wind power penetration levels, when energy storage prevents wasted wind. The same is the case for an extra interconnector to Norway operated as assumed here. Notably, the use of heat boilers not only saves operating cost but also CO2 emissions. Since the use of heat boilers at CHP locations specically tackles the minimum-load problem as a result of CHP unit operating constraints, heat

Figure 11 CO2 emission savings by exible CHP units, energy storage options and extra interconnection to Norway, no international exchange 290 & The Institution of Engineering and Technology 2009 IET Renew. Power Gener., 2009, Vol. 3, Iss. 3, pp. 279 292 doi: 10.1049/iet-rpg.2008.0080

Authorized licensed use limited to: Edinburgh University. Downloaded on January 11, 2010 at 06:55 from IEEE Xplore. Restrictions apply.

www.ietdl.org
boilers reduce the amount of wasted wind. Since the CO2 emissions of boilers and wind power are lower than CO2 emissions of CHP units, boilers reduce the overall amount of CO2 emitted by the system as well. possibilities for international exchange exist, wind power signicantly reduces imports and increases exports of the area it is integrated into. In the case study performed here, it is shown that the presence of large-scale wind power in Germany limits the useof exports for wind power integration in the Netherlands during some periods. Still, international exchange is shown to be key for wind power integration, especially at high penetration levels. As such, possibilities for international exchange should be regarded as a promising alternative for the development of energy storage in the Netherlands itself.

Summary and conclusions

A previously applied UC ED tool is extended to comprise the North-West European interconnected system. Representative models have been developed of the Netherlands power system, the neighbouring areas and the interconnections between these. Annual simulations have been performed for a range of wind power penetrations of 0 12 GW in the Netherlands, market designs (isolated system exible use of interconnections). Technical limits to the system integration of wind power in the Dutch system have been identied and the economical and environmental impacts of wind power on system operation quantied. Furthermore, the opportunities of energy storage and heat boilers for the integration of wind power in the Dutch system have been explored. Pumped accumulation energy storage (PAC), UPAC, CAES, the use of heat boilers at selected CHP locations and increased interconnection capacity with Norway (NN2) may provide additional technical space for wind power integration. The simulation results indicate that for the Dutch thermal generation system, ramp rate problems due to the aggregated variations of load and wind power are rare. This can be explained by the existing commitment constraints imposed on base-load coal units (must-run status) and CHP units due to heat demand, resulting in high operating reserve levels. The high reserve levels provide sufcient ramping capacity for balancing wind power variability in addition to existing load variations. For the optimisation of system operation with large-scale wind power, it can be noted that accurate, actualisations of wind power output and a continuous re-calculation of UC ED are essential. Although the additional variations introduced by wind power can be integrated, limits for wind power integration increasingly occur during high wind and low load periods. Depending on the international market design, signicant wind power opportunity may have to be wasted to prevent minimum-load problems. Wind power integration benets from postponed gate closure times of international markets, as international exchange may be optimised further when improved wind power predictions become available. The simulation results show that wind power production reduces total system operating cost, mainly by saving fuel cost. Wind power reduces the number of full-load hours of base-load coal-red generation, and to a lesser extent those of CCGT (with and without CHP-function). This has particular impacts on the prots of owners of these conventional generation units. By replacing fossil-red generation, wind power signicantly reduces the total exhaust of emissions (CO2 , SO2 , NOx). In case IET Renew. Power Gener., 2009, Vol. 3, Iss. 3, pp. 279 292 doi: 10.1049/iet-rpg.2008.0080

Acknowledgments

This work is part of the project PhD@Sea, which is funded under the BSIK programme of the Dutch Government and supported by the consortium We@Sea, http://www.we-atsea.org. The authors acknowledge the use of system data from Dutch TSO TenneT. Mr. Uwe Zimmermann and Mr. Friedhelm Witte of EON Energie, Dr. Bernhard Ernst of RWE and Mr. Stephan Schlunke of Vattenfall Europe are acknowledged for providing the German wind power data for their respective control zones.

References

[1] CARRION M. , ARROYO J.M.: A computationally efcient mixed-integer linear formulation for the thermal unit commitment problem, IEEE Trans. Power Syst., 2006, 21, (3), pp. 1371 1378 [2] PEDERSEN J. , ERIKSEN P.B.: Simulation model including stochastic behaviour of wind. Proc. Fourth Int. Workshop on Large-Scale Integration of Wind Power and Transmission Networks for Offshore Wind Farms, Billund, Denmark, 20 21 October 2003, p. 5 [3] HOLTTINEN H.: The impact of large scale wind power production on the Nordic electricity system. PhD Thesis, Helsinki University of Technology, Finland, 2004 [4] MEIBOM P., LARSEN H.V., BARTH R., BRAND H., WEBER C., VOLL O.: Wilmar joint market model documentation, Ris-R-1552. Technical Report, 2004, p. 60 [Ris National Laboratory] [Online], available: http://www.wilmar.risoe.dk/ [5]
MEIBOM P., KIVILUOMA J., BARTH R., BRAND H., WEBER C., LARSEN Value of electric heat boilers and heat pumps for wind power integration, Wind Energy, 2007, 10 , pp. 321 337 H.V. :

[6] SWIDER D.J.: Compressed air energy storage in an electricity system with signicant wind power generation, IEEE Trans. Energy Convers., 2007, 22, (1), pp. 95 102 [7] Operation Simulation Associates, Inc. PowrSym3 Users Manual v361d2 Technical Report 291

& The Institution of Engineering and Technology 2009

Authorized licensed use limited to: Edinburgh University. Downloaded on January 11, 2010 at 06:55 from IEEE Xplore. Restrictions apply.

www.ietdl.org
[8] UMMELS B.C., GIBESCU M., PELGRUM E., KLING W.L., BRAND A.J.: Impacts of wind power on thermal generation unit commitment and dispatch, IEEE Trans. Energy Convers., 2007, 22, (1), pp. 44 51 [9] UMMELS B.C., PELGRUM E., KLING W.: Integration of large-scale wind power and use of energy storage in the Netherlands, IET Renew. Power Gener., 2008, 2, (1), pp. 34 46 [10] UMMELS B.C.: Power system operation with large-scale wind power in liberalised environments. PhD Thesis, Delft University of Technology, The Netherlands, 2009 [11] TenneT TSO BV: Quality and capacity plan 20082014. Technical Report, 2008, available online at http://www. tennet.org/english/, last accessed October 2008 [12] ETSO: Net transfer capacities matrix, 2008, available online at http://www.etso-net.org/, last accessed October 2008 [13] GIBESCU M., BRAND A.J., KLING W.L.: Estimation of variability and predictability of large-scale wind energy in the Netherlands, Wind Energy, no. published online, p. 20, 29 September 2008 [14] DIMITROVSKI A., TOMSOVIC K.: Impact of wind generation uncertainty on generating capacity adequacy. Proc. Ninth Int. Conf. Probabilistic Methods Applied to Power Systems, Stockholm, Sweden, 11 15 June 2006, p. 6 [15] VAN WIJK A.J.M.: Wind energy and electricity production. PhD dissertation, Universiteit Utrecht, The Netherlands, 1990 [16] Union for the Co-ordination of Transmission of Electricity: Consumption data available online at http:// www.ucte.org/resources/dataportal/, last accessed October 2008 [17] UCTE: System adequacy forecast 2008 2020. Technical Report, 2008, available online at http://www. ucte.org/resources/publications/systemadequacy/, last accessed October 2008 [18] TenneT TSO BV: Vision 2030. Technical Report, 2008, available online at http://www.tennet.org/english/, last accessed October 2008 [19] National Grid: available online at http://www. nationalgrid.com/uk/electricity/sys/, last accessed October 2008 [20] European Commission Eurostat Homepage: available online at http://epp.eurostat.ec.europa.eu/, last accessed October 2008 [21] European Wind Energy Association: available online at http://www.ewea.org/, last accessed October 2008 [22] Transition Platform Sustainable Electricity Supply: Onderzoek naar de Toegevoegde Waarde van Grootschalige Elektriciteitsopslag in Nederland (in Dutch). Technical Report, 2008 [23] International Energy Agency: World Energy Outlook 2008. 2008, available online at http://www.worldenergyoutlook.org/, last accessed October 2008

292 & The Institution of Engineering and Technology 2009

IET Renew. Power Gener., 2009, Vol. 3, Iss. 3, pp. 279 292 doi: 10.1049/iet-rpg.2008.0080

Authorized licensed use limited to: Edinburgh University. Downloaded on January 11, 2010 at 06:55 from IEEE Xplore. Restrictions apply.